Meteorological and Oceanographic Ensemble Forecast Application System (EFAS)
The Ensemble Forecast Application System (EFAS) is a post processing engine that exposes the information rich ensemble data to forecasters and on-scene operators, and to existing Tactical Decision Aids (TDAs).
Ensembles of oceanographic and meteorological numerical forecast models are an excellent tool for quantifying the uncertainties in the marine environment that impact tactical operations. One challenge is to distill the information into a look and feel that the forecaster and operator quickly understand and to format it so that it can be put into TDAs. EFAS distills ensembles into a the familiar look and feel of a deterministic forecast for the operator, and it fits into their existing TDAs. Our approach is to use the probabilistic information contained in the ensemble to improve forecast skill and guidance as it applies to the specific operation being supported. EFAS first applies a bias-correction to some of the ensemble parameters to improve their forecast skill. For example; we see that forecasts of temperature, pressure, wave height, and wind speed are improved by applying bias-corrections, while wind direction forecasts are not. Then by using a consensus finding algorithm based on the RMSE history of the forecast parameter, it is possible to select the most skillful forecast value or forecast field or member from the ensemble. From the ensemble members one can also extract a spread around that forecast value. This spread, based on the operator’s requirement for accuracy or the operator’s tolerance in the forecast error, can also be used to state the confidence (high or low) that the forecast will meet the operator’s needs as it pertains to the operation. The ensemble generation is accomplished at any large central computing sites.
The EFAS post processing, which aggregates and distills the ensembles for use, can be at any location provided EFAS has access to the raw ensemble data sets. The on-scene operator interfaces the EFAS output which has the look and feel of a deterministic forecast, to derive the specific information needed to support their operation.